Testing is one of the most important phases during the realization of a software project, because it allows to identify the differences between the observed and expected output. However, often tests classes (generated both manually and automatically) have a lack of quality, that can be caused by the presence of "Test smells", a non-optimal design choices in the implementation of test code. As reported by recent studies, the presence of Test Smell could not only negatively influence the understandability of test suites, but also lead to less-effective test cases in finding bugs within the production code. The manual generation of test cases is a very expensive operation in terms of time, so a number of tools for the automatic generation of test cases have been proposed (e.g., EvoSuite). These tools, that at the beginning only focused on generating test cases having a high code coverage, used to generate low-quality test cases. For this reason, they were extended in order to allow the generation of test cases with both good coverage and quality (e.g., MOSA, QMOSA).
The aim of this thesis is to provide an automatic test case generation algorithm that is able to generate test cases that will not contain test smells (or at least only in a limited number). In the context of this study, we focused on identifying six specific tests mells: Eager Test, Mystery Guest, Resource Optimistism, Fire And Forget, Conditional Test Logic, Indirect testing.

"SmellIntensityMOSA" and "NoSmellMOSA" are the names of the approaches that I realized. These two approaches, extends MOSA and they are implemented into EvoSuite. I realized this work with Java on Intellij.

It's possible to execute these two approaches using EvoSuite.